In this section, we briefly give the overview the IEEE 802.11n protocol that we have used for throughput measurements and implementations of the elastic WLAN system. IEEE 802.11 proto-cols is a group of the standards created by the IEEE 802 LAN/MAN (Local Area Network/ Metropoli-tan Area Network) SMetropoli-tandards Committee. They specify over-the-air interfaces between a wireless client and a base station or between two wireless clients within a local area in either fixed, portable, or moving stations mode [16].
IEEE 802.11n is an amendment to the IEEE 802.11-2007 wireless networking standard. This standard was introduced with the 40 MHz bandwidth channel calledchannel bonding, the Multiple-Input-Multiple-Output (MIMO), the frame aggregation, and security enhancements over its prede-cessors. Table 2.1 briefly summarizes the IEEE 802.11n protocol.
Table 2.1: IEEE 802.11n specification.
specification IEEE 802.11n
frequency band 2.4 GHz 5 GHz
number of available channels 13 19 number of uninterfered channels 2 9
maximum speed 600Mbps
maximum bandwidth 40 MHz
number of maximum streams 4
maximum modulation 64 QAM
In the channel bonding technique, two adjacent channels within a given frequency band are combined together to increase the transmission speed. In the IEEE 802.11n protocol, the physical data rate can be doubled by using two adjacent 20 MHz channels simultaneously.
The spatial multiplexing and the space-time block coding are two MIMO-specific innovations of the IEEE 802.11n. For the spatial multiplexing, the transmitter transmits multiple independent data streams simultaneously from multiple antennas to increase the data rate. For the space-time block coding, the transmitter transmits dependent data stream which is spatially and time encoded to increase the signal reliability. Table 2.2 shows the throughput enhancement of IEEE 802.11n protocol using channel bonding and MIMO.
Table 2.2: Effects of channel bonding and MIMO on IEEE 802.11n throughput.
Stream number Bandwidth
20 MHz 40 MHz
1 stream 72.2Mbps 150Mbps
2 streams 144.4Mbps 300Mbps
3 streams 216.7Mbps 450Mbps
4 streams 288.9Mbps 600Mbps
IEEE 802.11n also provides the performance improvement through the frame aggregation in the MAC layer. The frame aggregation can transmit multiple frames as one big frame with a single pre-ample and header information to reduce the overhead by them. IEEE 802.11n introduces Ag-gregation of MAC Service Data Units (A-MSDUs) and AgAg-gregation of MAC Protocol Data Units
(A-MPDUs). The frame aggregation is a process of packing multiple A-MSDUs and A-MPDUs together to reduce the overheads and average them over multiple frames, thereby increasing the user level data rate [17].
2.2.1 Link Speed Change Feature of IEEE802.11n
The link speed or throughput is affected by many factors such as the modulation and the coding scheme, the transmission power, the transmission distance, and even the design of the network adapters [18]. Therefore, the theoretical computation of the link speed is very difficult. In this thesis, we take an alternative approach of conducting real-world measurements to model the actual link speed.
• Link Speed Model
Through measurements [19], we derived the following link speed (Mbps) function f(x) with the transmission distance x(m).
Table 2.3: Devices and software for measurements of IEEE 802.11n link speed.
model Ultra book Lesance NB S3441/L PC1 CPU Intel(R) Core i5 3317U (2.6 GHz)
OS Windows 7
model Ultra book Lesance NB S3532-SP PC2 CPU Intel(R) Core i3 2350M (2.3 GHz)
OS Windows 7
NIC Buffalo WZR-G1750DHP
software iperf2.0.5
protocol TCP
f(x)=
−2.20×10−3x3+1.85×10−1x2−5.33x+117 if 0≤ x<40
−6.00×10−4x3+9.50×10−3x2−1.73x+117 if 40≤ x<75 4.38×10−4x3−1.10×10−1x2+8.48x−189 if 75≤ x< 100 1.0 ifx≥100
(2.1)
The distance between theith host and the jth AP is defined by the Euclidean distance:
d(i, j)= q
(xhi − xAPj )2+(yhi −yAPj )2 (2.2) where xhi,yhi,xAPj , andyAPj are thexandycoordinate of theith host and the jth AP.
For our measurements, two PCs with the setting listed in Table 2.3 are prepared respectively as the source and destination nodes. The link speed is measured when both end nodes adopt the IEEE 802.11n protocol. At the measurements, the link distance between two end nodes is increased from 1m to 110m with the 5m interval. The parameters in iperfare set at 50 seconds for the measurement time, and 8Kbytes and 477Kbytesfor the buffer size and the window size respectively. Figure 2.4 shows the link speed measurement and estimation results.
0 10 20 30 40 50 60 70 80 90 100 110 0
10 20 30 40 50 60 70 80 90 100 110 120
Distance (m)
Linkspeed(Mbps)
measurement
approximation in Eq. (2.1)
Figure 2.4: IEEE 802.11n link speed measurement results.
According to our measurements, the peak throughput becomes about 115Mbpsat 1m dis-tance. Then, the effective throughput is rapidly dropped as the link distance increases. The throughput becomes the half of the peak at 40mdistance, and is about 10Mbpsat 80m dis-tance.
• Wall Effect
Since a WLAN system is based on radio frequency (RF) signals, factors affecting RF signal strengths should be taken into considerations for planning the efficient WLAN system. As the RF signal at 2.4GHzor 5GHzbecomes weak after penetrating obstacles such as concrete walls in a building, the throughput becomes low, which is calledwall effectin this paper. The type of the materials used in the obstacle determines the drop rate of the link speed. Besides, mobile operators (carriers) offer different mobile data plans to meet various subscribers demands and satisfactions. Depending on a plan, several choices can be taken for the link speed to connect to the Internet, which is called theMAP speed in this paper. To make the WLAN adaptable to such situations, these factors must be considered.
To examine the wall effect, we measured throughputs between a host and an AP that are located in different rooms separated by one concrete wall. The distance is fixed at 5m. Then, it is confirmed that the link speed is dropped by about 15% when the signal passes through one concrete wall in a room. In simulations, the link speed is decreased by multiplying 85/100 if there exists a wall between an AP and a host. Figure 2.5 shows the speed drop by the wall effect.
• Device Effect
Furthermore, it is found that a PC provides a maximum of 54Mbpswhen it works as a VAP while a MAP supports a maximum of 30Mbps speed. Thus, in this thesis, the link speed estimated from the link speed function f(x) is adjusted by multiplying 45/100 for the VAP
51 Mbps
60 Mbps 15% speed drop
Access Point
Concrete Wall
Host
Host
Figure 2.5: Wall effect.
(a) Dedicated AP (DAP) (b) Virtual AP (VAP) (c) Mobile AP (MAP)
Figure 2.6: Three types of APs.
and 25/100 for the MAP. The same wall effect is applied for any AP type. These parameters are used in our algorithm and theWIMNET simulatorfor simulations.